def test_standalone_rpca(self):
"""Checks the output produced by gemelli's standalone script.
This is more of an "integration test" than a unit test -- the
details of the algorithm used by the standalone CTF script are
checked in more detail in gemelli/tests/test_factorization.py.
"""
in_table = get_data_path('test-small.biom')
in_meta = get_data_path('test-small.tsv')
out_ = os_path_sep.join(in_table.split(os_path_sep)[:-1])
runner = CliRunner()
result = runner.invoke(standalone_ctf,
['--in-biom',
in_table,
'--sample-metadata-file',
in_meta,
'--individual-id-column',
'host_subject_id',
'--state-column-1',
'context',
'--output-dir',
out_])
# check exit code was 0 (indicating success)
CliTestCase().assertExitCode(0, result)
# Read the results
samp_res = pd.read_csv(
get_data_path('context-subject-ordination.tsv'),
sep='\t',
index_col=0)
feat_res = pd.read_csv(
get_data_path('context-features-ordination.tsv'),
sep='\t',
index_col=0)
# Read the expected results
samp_exp = pd.read_csv(
get_data_path('expected-context-subject-ordination.tsv'),
sep='\t',
index_col=0)
feat_exp = pd.read_csv(
get_data_path('expected-context-features-ordination.tsv'),
sep='\t',
index_col=0)
# Check that the distance matrix matches our expectations
comp_col = ['PC1', 'PC2', 'PC3']
cent_ = samp_res[comp_col].mean().values.max()
self.assertAlmostEqual(cent_, 0)
cent_ = feat_res[comp_col].mean().values.max()
self.assertAlmostEqual(cent_, 0)
# check matched
assert_allclose(absolute_sort(samp_res[comp_col].values),
absolute_sort(samp_exp[comp_col].values),
atol=.5)
assert_allclose(absolute_sort(feat_res[comp_col].values),
absolute_sort(feat_exp[comp_col].values),
atol=.5)
def test_qiime2_ctf(self):
"""Tests that the Q2 and standalone ctf results match.
Also validates against ground truth "expected" results.
"""
# Run gemelli through QIIME 2 (specifically, the Artifact API)
res = q2gemelli.actions.ctf(table=self.q2table,
sample_metadata=self.q2meta,
individual_id_column=self.subj,
state_column=self.state)
oqza, dqza, sqza, fqza = res
# Get the underlying data from these artifacts
q2straj = sqza.view(pd.DataFrame)
q2ftraj = fqza.view(pd.DataFrame)
# Next, run gemelli outside of QIIME 2. We're gonna check that
# everything matches up.
# ...First, though, we need to write the contents of self.q2table to a
# BIOM file, so gemelli can understand it.
# Derived from a line in test_standalone_ctf()
out_ = os_path_sep.join(self.in_table.split(os_path_sep)[:-1])
# Run gemelli outside of QIIME 2...
runner = CliRunner()
result = runner.invoke(standalone_ctf,
['--in-biom',
self.in_table,
'--sample-metadata-file',
self.in_meta,
'--individual-id-column',
'host_subject_id',
'--state-column-1',
'context',
'--output-dir',
out_])
# check exit code was 0 (indicating success)
self.assertEqual(result.exit_code, 0)
# ...and read in the resulting output files. This code was derived from
# test_standalone_ctf() elsewhere in gemelli's codebase.
samp_res = read_csv(
get_data_path('context-subject-ordination.tsv'),
sep='\t',
index_col=0)
feat_res = read_csv(
get_data_path('context-features-ordination.tsv'),
sep='\t',
index_col=0)
# Check that the trajectory matrix matches our expectations
comp_col = ['PC1', 'PC2', 'PC3']
assert_allclose(absolute_sort(samp_res[comp_col].values),
absolute_sort(q2straj[comp_col].values),
atol=.5)
assert_allclose(absolute_sort(feat_res[comp_col].values),
absolute_sort(q2ftraj[comp_col].values),
atol=.5)